ZK proofs enable low-latency, trustless, zero-overhead bridging between permissionless and sovereign ZK chains with mixed data availability models. We call it hyperbridging. Not necessarily atomic, but seamless: one single confirmation in Metamask will let you interact with any smart contract on any other L2 in the ZK network, sending the result back to your wallet, in seconds. This is a prerequisite for limitless scalability on Ethereum, and will solidify it as the global settlement layer of the Internet of Value. Let's break it down: - Low-latency. With sufficient parallelization, we can bring down the proof generation time for a block to seconds. The actual latency will be determined by the users preference for finality guarantees (either Ethereum will embrace single-slot finality, or chains will rely on economic finality guarantees), but it will be low in any case (seconds). But even if you insist on waiting to reach full finality on L1 under today's model, all the other benefits hold and are not possible with anything but ZK rollups. - Trustless. Every user verifies the validity of all of the transactions on all chains and all the bridging transactions by relying purely on math. You never need to trust any validator, watcher, relayer—anyone else but yourself. - Zero-overhead. With state diff architectures (uniquely enabled by ZK rollups and currently embraced by zkSync and Starknet), a bridging tx will cost nearly as much as a normal tx on either chains; if the chains have different costs, e.g. because one account is a rollup and the other one is a validium, total_cost = tx_cost(chain A)/2 + tx_cost(chain B)/2. This is because data availability is our main cost driver, and with state diffs you only pay for the storage slots you update where you update them. - Permissionless, sovereign ZK chains. Ethereum is so powerful because anyone can deploy a smart contract. You don't have to ask Vitalik for permission or sign a contract with the Ethereum Foundation. To reach hyperscale, the same must be true for deploying L2s, an unlimited number of them. But that means most other L2s cannot be trusted, by default. No one has the capacity to verify all the value transactions of the world. Thus, most bridging must rely on fully trustless mechanisms. Using ZK proofs, it's easy. A single proof on L1 can attest to the validity of all the thousands of ZK rollups—and, what's very important, validiums (which will be responsible for the bulk of all transactions thanks to much lower costs). For optimistic rollups, you either have to wait 7 days, or you must explicitly whitelist a small number of shared sequenced rollups, which kills the ability to deploy L2s permissionlessly. - Mixed data availability (DA) models. Rollups will always be relatively expensive, by design (because, unlike sidechains and alt L1s, they trustless prioritize security and censorship-resistance). Most L2s will thus have to use alternative DA layers which are much cheaper. The beauty of validiums and volitions (such as zkSync’s zkPorter) is that this affects neither verifiability nor bridgeability at zero overhead. The same does not hold true for optimistic rollups, though. There is simply no way to trustlessly bridge between an optimistic rollup and an Anytrust chain (such as Arbitrum Nova), even with a shared sequencer: by mixing optimistic DA models, or you will automatically downgrade all the optimistic rollups in this shared sequencing scheme to the least secure of them, rendering them not much more secure than a sidechain. Nothing has changed about these fundamentals. Ethereum will win not even because of its first-mover advantage, but because its ethos and culture is focusing us on solving the right problems. What I describe above gives you infinite scalability – not 10k, not 100k, but arbitrary many TPS and users – while not sacrificing UX, decentralization, trustlessness, or resilience in front of mighty adversaries. Have just a little patience, you will soon see all of it in action. I promise. 🍿